Fault Tolerant Deep Neural Networks for Detection of Unrecognizable Situations

被引：5

作者：

Rhazali, Kaoutar ^{[1
,2
]}

Lussier, Benjamin ^{[1
]}

Schon, Walter ^{[1
]}

Geronimi, Stephane ^{[2
]}

机构：

[1] Univ Technol Compiegne, Sorbonne Univ, CNRS, UMR 7253, Heudiasyc CS 60 319, F-60203 Compiegne, France

[2] Direct Rech & Innovat Automobile, Grp PSA, Route Gisy 78943, F-78943 Velizy Villacoublay, France

来源：

IFAC PAPERSONLINE | 2018年 / 51卷 / 24期

关键词：

Safety; Fault Tolerance; Artificial Intelligence; Neural Networks; Autonomous Vehicles;

D O I：

10.1016/j.ifacol.2018.09.525

中图分类号：

TP [自动化技术、计算机技术];

学科分类号：

0812 ;

摘要：

Deep Neural Networks are achieving great success in various fields. However, their use remains limited to non critical applications because their behavior is unpredictable and unsafe. In this paper we propose some fault tolerant approaches based on diversifying learning in order to improve DNNs dependability and particularly safety. Our main goal is to increase trust in the outcome of deep learning mechanisms by recognizing the unlearned inputs and preventing misclassification. (C) 2018, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved.

引用

页码：31 / 37

页数：7